The Impact of Generative AI-Powered Recommendation Systems on Purchase Intention of Virtual Sports Consumers

To explore the unique psychological mechanisms through which generative AI recommendation systems influence consumers' purchase intentions, this study employed a two-stage experimental design, combined with Structural Equation Modeling (SEM) and fuzzy-set Qualitative Comparative Analysis (fsQCA), to conduct an in-depth investigation of the virtual sports consumption context. The research findings are as follows: 1) The effectiveness of the experimental manipulation was confirmed: compared to traditional systems, generative AI recommendation systems successfully induced significantly higher perceptions of anthropomorphism among users. 2) SEM analysis revealed a critical chain mediation path: system type positively influences users' sense of social presence, which in turn enhances their perceived trust, and ultimately has a significant positive impact on purchase intention. 3) To further clarify the multiple concurrent paths leading to high purchase intention, fsQCA analysis identified six effective configurations, which can be summarized into four modes: (1) a "dual-engine" mode emphasizing both trust and presence; (2) an "innovative dualism" mode targeting innovative consumers, where trust or presence plays a compensatory role; (3) an "engagement-innovation synergy" mode predicated on high involvement and innovativeness; and (4) an "all-factor synergy" mode integrating all positive conditions. The study reveals the complex mechanisms by which generative AI influences consumer decision-making through a "social-emotional" pathway, providing empirical evidence for the theoretical development and practical design of next-generation recommendation systems.